CN102364962A - A Multipath Transmission Scheme Based on Scalable Layered Coding - Google Patents

Abstract

The invention relates to a scalable layered coding based multi-path transmission scheme, which comprises the following steps that: a sink node builds an energy time delay gradient ET and a time delay gradient T for each node; a video sequence is subjected to M-class scalable coding, and a bit rate of each layer after coding is ...., (wherein <K2 <...<), different marks are set for a basic layer and enhancement layers, P represents a channel bandwidth, if, then the base layer is transmitted only; if (1<q<M), then first q bitstream layers are transmitted, if, then M bitstream layers are transmitted; a first data packet in the bitstream layer to be transmitted is selected to be used as a detection data packet to be sent to the sink node; the sink node receives the detection data packet to carry out reverse path enhancement; and the subsequent bitstream layers select paths that the data packets with the same mark value pass to be used as transmission paths to send the data to the sink node. By using the scheme provided by the invention, the change in network bandwidth can be adapted, and the business requirements of different bandwidth users can be met.

Description

A Multipath Transmission Scheme Based on Scalable Layered Coding

technical field

The invention is a multi-path transmission scheme proposed for the code stream after scalable layered coding of video sources in the wireless multimedia sensor network. It is mainly used to solve the transmission problem of video information in a bandwidth-limited network, and belongs to the technical field of transmission in a wireless multimedia sensor network. the

the

Background technique

In Wireless Multimedia Sensor Networks (WMSN), the energy of nodes is limited, the network bandwidth is lower than that of wired channels, and there are many factors such as signal attenuation and noise interference, which make video stream transmission face the problem of packet delay. Large-scale fluctuations, losses, flow restrictions, etc. To address these issues, it is necessary to study video encoding and transmission methods. Among the current coding technologies that are more suitable for real-time streaming media transmission are multi-description coding and layered coding, and multi-description coding is to generate multiple equally important code streams from the video source, so the code stream cannot adapt to the network bandwidth and receiving equipment. The characteristics are adaptively adjusted. For video transmission, multi-path routing is widely used at present, and at the same time, it is necessary to combine the transmitted code streams to select suitable routes for the code streams with different needs. the

The scalable layered coding method can adapt to the change of network bandwidth and satisfy user terminals with different processing capabilities. In practical applications, even if the code stream of the enhancement layer is lost, the receiving end can still maintain watchable video quality. Therefore, scalable layered video coding can well support networks with different QoS methods such as differential services. the

The bandwidth resources in WMSN are limited, and the transmission of video streams has a large transmission volume and high real-time requirements. The strategy of multi-path transmission is proposed to meet this characteristic. There are two modes of existing multi-path transmission: one is master/standby mode, data is transmitted randomly along a path, or along a pre-defined path, when the main path fails, the backup path can take over Transmission tasks; Another multi-path transmission technology is the replication mode, where the source node replicates the data and transmits along all the paths at the same time to improve redundancy. Although these two methods have certain effects, the former limits the potential of the backup path and will quickly consume the energy of the main path; the latter increases data redundancy and wastes resources and energy, and neither is effective Combining with the characteristics of the code stream after the video source has undergone scalable encoding, there is no suitable route for different code streams. the

Contents of the invention

Technical problem: The purpose of this invention is to provide a multi-path transmission scheme in WMSN to adapt to the code stream after the video source has undergone scalable layered coding, so as to solve the problem of video information transmission in the bandwidth-limited network.

Technical solution: The method of the present invention is a strategic method. By modifying the directional diffusion routing protocol in WMSN, an energy delay gradient and a delay gradient are established for each node, and multiple nodes are established for the source node and the sink node according to the gradient information. Paths with different quality of service, so that each layer of code stream after the video source can be scalable and encoded can be selected and transmitted more reasonably according to its own characteristics.

1. Architecture

    The basic principle of video transmission using scalable layered coding is shown in Figure 1, that is, video data with a certain resolution generated by a certain source is cumulatively coded, and divided into several domains such as time domain or space domain. Layers, including a base layer and one or more enhancement layers, each layer data can be transmitted independently after encoding. The base layer can be decoded independently, while the enhancement layer code stream needs to be decoded together with the base layer. After being decoded by a certain decoder, the base layer code stream can guarantee the most basic video quality, while the enhancement layer code stream can provide enhanced details on the video quality. The basic layer contains the most basic video information and key time information, called key frames, which require high reliability and real-time performance in the transmission process; the enhancement layer contains detailed information in the video frame, which is the video coding The optional information of the frame is called a normal frame, which has higher requirements on real-time performance and lower requirements on reliability. Therefore, different paths can be selected according to different requirements in the transmission process of the two code streams.

The process of establishing multiple paths in WMSN network with multipath transmission based on directed diffusion routing protocol includes three stages: interest diffusion, gradient establishment and path strengthening. In the gradient establishment stage, the traditional directed diffusion routing protocol only establishes one gradient item for each node, so that the established multiple paths cannot distinguish the path service quality. The improved method of this paper establishes two gradient items for each node , that is, the energy delay gradient and the delay gradient (referred to as ET and T, respectively). In terms of multipath construction, the traditional directed diffusion routing protocol only needs to construct multiple paths with one quality of service, while the improved method in this paper needs to construct paths with multiple different qualities of service. the

The following is a specific description of the improvement of the two aspects of gradient establishment and multipath construction in the directed diffusion routing protocol. the

Gradient establishment: When the aggregation node floods interest messages to the entire network, the gradient is established. Each node maintains an interest record table. For each interest received by the node, there is an entry in the list to record the sending ET and T of the neighbor nodes of the interest message.

The establishment method of ET and T is: for a node in the sensor network, if it receives n interest messages from m neighbor nodes (discarding the same interest messages from the same neighbor node), first, according to the node The sequence of arrival of received messages of the same interest establishes a delay gradient. The interest message that arrives first has the largest neighbor node delay gradient T, which is set to G (G should be much larger than n). For the i-th arriving interest message, set the gradient value of its neighbor node to T=G-i +1; Secondly, since each interest message carries the remaining energy of neighbor nodes, the remaining energy of m different neighbor nodes sending the same interest message is sorted from large to small. If the ranking is j, the last ET=T-j. In the wireless multimedia network shown in Figure 2, if node 6 receives n=6 interest messages from m=5 neighbor nodes, the interest messages from neighbor nodes 1, 2, 3, and 4 The message is interestA, the arrival order is 1, 2, 3, 4; the interest message sent by neighbor nodes 4 and 5 is interestB, the arrival order is 1, 2, let G=100, and the remaining energy of each neighbor node is E1=4J , E2=3J, E3=1J, E4=2J, E5=5J, then the interest gradient table after ET and T are established is shown in Figure 3. 

Multi-path construction: After the gradient is established, when a sensor node collects data that matches the interest message in its interest table, it indicates that the data is the data required by the sink node, and the node sends a detection packet to the sink node. Build the right path.

Assuming that the source node needs to send N detection data packets, each detection data packet has a packet header identifier to distinguish it, which is set to 0, 1, ..., N-1, and the process of constructing a multipath for the detection data packet is: source The node selects different next-hop nodes for N detection packets with different header identifiers according to the following node selection criteria, and the next-hop node continues to forward according to the same node selection criteria until it reaches the sink node, thus forming a network from the source node to the sink node. N paths of sink nodes. the

Next-hop node selection criteria: check the interest gradient table of node A, find the entry corresponding to the interest message that matches the data to be transmitted, and select the neighbor node B with the largest ET as the next jump node. If they have the same maximum value, select the neighbor node with the largest ET that appears first in the interest table; for the detection packet whose identifier is not 0, find out the first N-1 neighbor nodes with a larger ET except node B in the interest table value of neighbor nodes, randomly assign one as the next hop node. Taking the sensor network shown in Figure 3 as an example, suppose the source node collects the data corresponding to interestA in the interest gradient table and sends 4 detection data packets. When node 6 receives the detection data packets and forwards them, if the detection packets If the ID is 0, the neighbor node 1 is selected as the next-hop node; if the detection packet ID is not 0, the neighbor nodes 2, 3, and 4 are randomly assigned as the next-hop node. the

2. Method flow

The multi-path transmission method based on scalable layered coding is specifically: the source node establishes gradient information, performs scalable layered coding on the video source, determines the code stream layer to be transmitted, constructs multiple paths, and selects for different code stream layers. different paths for transmission. The specific process of this method is:

(1) The sink node floods interest messages to the entire network, and establishes gradients ET and T for each node.

(2) When a sensor node collects video data that matches its interest table, it performs M-level scalable encoding on the video source, and encodes it into a basic layer and M-1 enhancement layers. Set the bit rate of each layer after encoding Respectively, .... (where < <...). The sensor node sets different marks for the code stream of the base layer and the code stream of the enhancement layer, respectively, base layer: , enhancement layer 1: , ..., enhancement layer M-1: . Let the bandwidth of the channel be P, if , then only the basic layer will be transmitted; if (1<q<M), then the first q code stream layers will be transmitted; if , then the first M code stream layers will be transmitted. the

(3) If the first Q () code stream layers are to be transmitted, the first data packet of each layer in the Q layer is selected as the detection data packet to be sent, and each detection data packet selects a different next-hop node until it reaches A sink node that forms multiple paths from a source node to a sink node. the

(4) The convergence node strengthens the reverse path for the received Q data packets according to the path passed by each data packet, and establishes Q high-speed paths. the

(5) The subsequent basic layer code stream and enhancement layer code stream select the path through which the probe packet with the same tag value as its own passes, and send the data to the sink node. the

the

Beneficial effect

The invention proposes a solution to the problem of video information transmission in a bandwidth-limited environment in WMSN. When the bandwidth in the network is insufficient, it can at least ensure the efficient and reliable transmission of the basic layer and provide basic video service quality. As the bandwidth increases, more and more enhancement layers are received, and the video received by the receiving end The quality also increases until all stream layers are received and the video quality is at its best. Therefore, the transmission quality of the video in the bandwidth-limited network is guaranteed, and the requirements of users with different bandwidths can be adapted.

Description of drawings

Figure 1. The basic principle of video transmission using scalable layered coding. the

Figure 2 is a simulation diagram of a multimedia sensor network. the

Fig. 3 Interest gradient table after ET and T are established. the

Detailed ways

According to the scene diagram in Figure 2, it is assumed that the video data collected by node 6 needs to be sent to the sink node, the encoding method of the data is 4-level time-domain scalable coding, and the channel bandwidth is P, then the multi-path transmission of video data from node 6 to the sink node The process is:

1. The sink node floods interest messages to nodes 1, 2, 3, 4, 5, and 6. After each node receives the interest message, it creates an interest record table and records the T and ET of neighboring nodes.

2. Node 6 collects the data that matches its own interest record table, performs 4-level time-domain scalable encoding on the video source, encodes it into 1 basic layer and 3 enhancement layers, and adds a header identification to each data packet , Base layer: , Enhancement layer 1: , Enhancement layer 2: , Enhancement layer 3: , Let the bit rates of the base layer and enhancement layers 1, 2, and 3 after encoding be ,,, (where <<<). the

3. If , only the base layer is transmitted; if , the base layer and enhancement layer 1 are transmitted; if , the base layer and enhancement layers 1, 2 are transmitted; if , the base layer and enhancement layers 1, 2, 3 are transmitted. the

4. To transmit the Q() layer code stream, node 6 sends the first data packet of each layer code stream as a detection data packet, and the destination node is set as the sink node. the

5. Check the interest record table of node 6, find the entry corresponding to the interest message that matches the data to be transmitted, and select the node with the largest ET that appears first as the next hop node for the correct detection packet. , randomly assigning a node with a larger T value in the first Q-1 nodes except the node selected by the detection packet as the next-hop node. the

6. The next-hop node continues to forward according to the same method, and records the path of the packet in the data packet until it reaches the sink node. the

7. The sink node strengthens the reverse path for the received Q data packets according to the path recorded in each data packet, and establishes Q high-speed paths from node 6 to the sink node. the

8. Subsequent basic layer code stream and enhancement layer code stream select the path through which the detection packet with the same tag value as its own passes, and send the data to the sink node. the

Claims (3)

1. A multipath transmission scheme based on scalable layered coding, characterized in that it comprises the following steps: The aggregation node floods the interest message to the entire network, and establishes the gradient energy delay gradient ET and delay gradient T for each node; When a sensor node collects video data that matches its interest table, it performs M-level scalable encoding on the video source, and encodes it into a basic layer and M-1 enhancement layers. The bit rates of each layer after encoding are respectively ,…., (where<

<...<); The sensor node sets different marks for the code stream of the basic layer and the code stream of the enhancement layer, the basic layer: , the enhancement layer 1: , ..., the enhancement layer M-1: ; Let the bandwidth of the channel be P, if , then only the basic layer is transmitted; if (1<q<M), then the first q code stream layers are transmitted; if , then the first M code stream layers are transmitted; If you want to transmit the first Q () code stream layers, select the first data packet of each layer in this Q layer as the detection data packet to send, each detection data packet selects a different next-hop node until it reaches the sink node, Form multiple paths from source nodes to sink nodes; The sink node performs reverse path strengthening on the received Q data packets according to the path passed by each data packet, and establishes Q high-speed paths; The subsequent basic layer code stream and enhancement layer code stream select the path through which the probe packet with the same tag value as its own passes, and send the data to the sink node. 2. The multi-path transmission method based on scalable coding according to claim 1, characterized in that the step of (the) node establishing a gradient includes: establishing a delay gradient according to the order of arrival of the same interest messages received by the node, The interest message that arrives first has the largest neighbor node delay gradient T, which is set to G (G should be much larger than n). For the i-th arriving interest message, set the gradient value of its neighbor node to T=G-i +1; establish the energy delay gradient ET according to the remaining energy of the neighbor nodes carried in each interest message, and sort the remaining energy of m different neighbor nodes sending the same interest message from large to small, if the ranking is j, Then the final ET=T-j. 3. The multi-path transmission method based on scalable coding according to claim 1, characterized in that (the) step of selecting a next-hop node when detecting data packet transmission: check the interest gradient table of the node, and search for the data to be transmitted For the entry corresponding to the matching interest message, for the detection packet of the basic layer, select the neighbor node B with the largest ET as the next hop node; Neighbor nodes; for the rest of the detection packets, find out the first N-1 neighbor nodes with a larger T value except node B in the interest table, and randomly assign one as the next hop node.

Images